Isn’t llama.cpp already depending on Python anyways for the templating?

It uses a cpp implementation of jinja https://github.com/google/minja